Question Very high GPU hotspot temperature on 3080

[Espher_24]

Hi, I have a Suprim X 3080 gpu and after benchmarking Cyberpunk on it, the temps in HWinfo settled to around 74C max for the core and 72C max for the memory junction which are great.
However the hotspot went all the way up to 100C after some time. Is this reading accurate? The card has never been taken apart and was bought new.
Should I open it up? If so, do you recommend just repasting it or also changing the pads?
Thanks.

 

[parker_307]

it could be accurate. But how many times have you ran the test?

 

[Espher_24]

it could be accurate. But how many times have you ran the test?

I have run the Cyberpunk test 3-4 times back to back to see how high it would get and it reached 100.1C max.
With Furmak I've got similar results

 

[parker_307]

I have run the Cyberpunk test 3-4 times back to back to see how high it would get and it reached 100.1C max.
With Furmak I've got similar results

just did a bit of research and it seems to be a problem with the card. Id recommend repasting it. And tell me how that goes.

 

[Espher_24]

just did a bit of research and it seems to be a problem with the card. Id recommend repasting it. And tell me how that goes.

Do you mean a problem with the 3080 in general or with the MSI model? if the latter, can you give me your source? Would love to learn more. It sucks when you spend so much for a GPU and it comes like this...

 

[parker_307]

MSI model. View: https://www.reddit.com/r/overclocking/comments/mh722x/high_temps_on_a_msi_rtx_3080_suprim_x/

heres just one

 

[Espher_24]

MSI model. View: https://www.reddit.com/r/overclocking/comments/mh722x/high_temps_on_a_msi_rtx_3080_suprim_x/

heres just one

Oh wow, I even have the same case as the guy in the post. Guess it's repaste time... Will give it a shot in the next days. Thanks btw

 

[parker_307]

Oh wow, I even have the same case as the guy in the post. Guess it's repaste time... Will give it a shot in the next days. Thanks btw

yep. np. Mark as best answer if it fixes it. ty

 

[hotaru.hino]

Oh wow, I even have the same case as the guy in the post. Guess it's repaste time... Will give it a shot in the next days. Thanks btw

Make note of the spread of the old paste. If it looks unbalanced, it could also be an indication that something isn't letting the heat sink press evenly on the die.

 

[Karadjgne]

Just be aware that Furmark is absolutely brutal on a gpu. It is a known fact that Furmark (it even has a disclaimer) can and Has been responsible for gpu burnouts, it's potentially That brutal.

So while you might see extreme temp Hotspots, take that with a grain of salt, or three, as it's not really representative of anything close to 'normal' operation.

Instead, pick the most brutal game with brutal settings in game, like Ray Tracing with 4k DSR etc, totally max out everything possible, and then check that Hotspot. That'll be your personal limits for your card, with your usage, and anything lesser will be your 'normal'.

 

[Espher_24]

Small update, even with 100C hotspot the card did not thermal throttle and stayed above the 1905mhz boost clock for this model, averaging around 1950mhz with peaks to 1995mhz.
If the card didn't throttle does it mean that the hotspot reading isn't coming from the core?

 

[Phaaze88]

That's because hot spot doesn't throttle the core until it hits its maximum.
There wasn't really a problem to begin with.

 

[Espher_24]

That's because hot spot doesn't throttle the core until it hits its maximum.
There wasn't really a problem to begin with.

So you're saying that in the long run it's safe to run the card like this?

 

[Phaaze88]

So you're saying that in the long run it's safe to run the card like this?

The hot spot has a much higher temperature range than gpu core average does.
Gpu core max is 91C. Hot spot is at least 110C. Memory is 105C.
If it's just some spikes, they are a non-issue and don't count.